Re: bug in clsql MOP usage?

[Bruno H. <br...@cl...>]

Joerg wrote:
> >2. Even after an even number of definitions (no error
> >message), make-instance fails, trying to funcall the #<unbound> object.
> >There's then clsql-sys::view-class-effective-slot-definition
> >extraterrestrial> on the stack above #<subr clos::%initialize-instance>

This looks like an effective-slot-definition instance which is either
uninitialized or in a weird state. Can you show all the metaobject code
in clsql that deals with slot-definitions?

> >I haven't yet tested this with Bruno's yesterday (2004-10-06)
> >CVS changes. I'll do so ASAP.
>
> These changes only affected mop.tst, not a .lisp file.

Could you please try a more recent clisp CVS? I added more defensive
checking 3 days ago. I have a certain suspicion about what clsql is doing...

Bruno